GENERATIVE ARTIFICIAL INTELIGENCE: BETWEEN HYPE AND RESISTANCE

Artificial intelligence (AI) is not a new phenomenon, but its evolution throughout history has been modifying what we can understand for it.

Since the famous “Test de Turing” of the 1950s, in which it was considered that a machine was intelligent if when talking to a human by text, it made it so natural that that it was not able to distinguish that its interlocutor was a machine; even the current era of the peak of generating artificial intelligence, which can produce new content such as text, images, videos, voices, programming codes, etc.; we can say that its advancement and progress has been extraordinary.

All this is mainly possible by the exponential development of the ability to process and store large volumes of data.

Moreover, until just a few years ago, it was a discipline reserved to computer sciences, but by becoming cross-sectional to most activities and the everyday life of people, they began to engage in their study, analysis and experimentation, all professions, because none results alien, becoming an imperative that their approach is interdisciplinary.

In this current context: Are we aware of the revolution we are carrying on? Let's see:

The chatGPT boom

It is a conversational agent, which processes natural language, that is, can recognize and generate human language, through conversations with the user, which formulates a question, a request or provides instructions (Prompts), so that the system responds by text.

The more accurate and detailed the Prompts, better be the answer we get.

Difference of others chatbots for its great sophistication as it is within L, great language models, attentive to have been trained and rely on millions of text data parameters.

It is based on the model _ (pre-delivered generational transformer), developed by OpenAI, was evolving until it reached its 3.5 version, which boosted the launch of the ChatGPT product.

The really disruptive, which produced great social impact, was its appearance in the form of public and free access chatbot, thus democratizing its use, as it simply allows anyone to interact with the system by talking in a simple, natural and intuitive way, as if it did with another human.

This got its use massive in its first days, reaching the million users in just a week.

To this are added other of its main features such as the wide versatility, and the possibility of carrying out various tasks, which is extremely attractive since it can be used for numerous purposes and activities among which one can highlight:

• entertainment
• Education
• Travel guides and itineraries
• training plans
• Cooking recipes
• translations
• complex calculations
• document generation
• composition of stories, tales or poetry
• review of texts and extracting relevant information
• completion of abstracts
• draft, computer or trigger ideas

Recently, we have incorporated the possibility of creating our own custom GPTs, in a simple and no need to know how to write code, adapted to our specific needs, tasks and purposes.

This is done in the same way that we interact with the system to provide us with other types of responses, i.e. by giving you accurate instructions on what we want to create, comparing it with additional knowledge and skills available; those that can even be shared and made available by other users to use them.

Thus, the possibilities are numerous:

• customer service assistant
• generation and revision of specific documents
• content creation
• marketing strategies
• automation of tasks
• development of codes
• tutoring for study
• E- Mail management and schedule
• Documentation of meetings and ideas

The answers they provide are based on statistical probability of the information standards that extract from the data with which they have been trained, so the greater the quantity, quality and variety of the data, the more accurate the answers will be.

But they do not understand or have knowledge (in the human sense) about the concepts of the text they generate, for example, ChatGPT can draft a bill and perhaps do it perfectly, but do not know or understand laws.

They also do not understand the context of reality, nor have common sense, nor distinguish the true of the false, so their answers can count on bias, errors, inconsistencies or hallucinations (inventions).

This sums up the lack of transparency and security in the processing of personal data and the environmental impact that occurs by the amount of energy needed to train and use these models.

Therefore, the ethical and legal implications are numerous and often generate rejection or distrust in relation to their use and implementation in different sensitive areas such as education, justice or health.

And now: _

The announcement of its launch had a lot of repercussion, generated enthusiasm and high expectations, as the demonstration video shared by the company itself showed something spectacular and very overcoming: a real-time interaction by voice between the user and the system, with apparent autonomy and initiative on its own to see and analyze what was happening in the video.

But, less than twenty-four hours of his presentation had already generated disappointment, criticism and controversy, becoming that the shown was a placement on the scene, attentive that did not coincide with the company published in the developer article, which contained the explanation of how the results would be obtained.

In fact, fixed images and texts were used as Prompts to generate the answers, in a similar way to what can be achieved with GPT-4, which is not little, but it is much less than it seemed that it could offer or achieve.

In short, it was just a fake marketing or advertising strategy or at least deceptive to make it seem something that was not and achieve a greater repercussion, which ended up opacing the obvious potentiality with which this new development is regarded as very powerful and direct competitor of GPT-4.

We should wait a little longer to check.

Reactions and actions:

It is added that often the news is loaded with some sensationalism and other apocalyptic premonitions, that by the need for inmediacy and speed with which they circulate, we cannot even stop analyzing in detail and much less validate.

This can be overwhelming and for a moment we feel that we are within a series of science fiction, which generates an excessive expectation, fanaticism or even rejection of the fear of the losses that could be caused.

The irruption of ChatGPT has again put some statements that until recently were considered valid as to what AI can do, raising new doubts and debates about whether it has creativity or not, how and in what tasks or activities will replace us and whether it is possible that a general AI that equals human intelligence in an integral way or surpasses it (Supper intelligence).

Understanding in depth how these models work can be very complex.

By nature, humans are afraid of the unknown, and fear is even greater if they follow the belief that what we do not know can harm us or leave us without employment.

Focusing solely on highlighting your limitations, generates insecurity or confusion and leads to rejecting your implementation or use, or at least maintaining scepticism about you.

On the other hand, simplifying too much, can cause high expectations and an excess of optimism, which ends in frustration, making us fall into the error that for all we need AI and that always everything can do or solve properly.

In view of this, adopting a realistic posture allows us to focus on their true uses, to take advantage of all the benefits and advantages that can provide us with these tools that ultimately enhance human skills.

Being essential at the same time, to know its disadvantages, risks and limitations, knowing that AI cannot solve it all and not even it is necessary to use it in all cases, adopting measures to prevent improper use, possible damage and always maintaining control and decision making in the position of human beings (human in the loop).

But this is also not easy, as it requires knowledge and this can only be acquired through education, access to quality information, training and constant learning, which allows to develop critical thinking, adaptation to changes and flexibility, human skills that we must strengthen and deepen, being so necessary in the present as technology.

The popularization of ChatGPT is a clear example that we are all invited to engage and actively participate in this era of the IA generation.

What will come:

What we think now will probably change, since it will also become our relationship with AI and we cannot know what will happen.

Without prejudice to this, one can see that the generational AI has come to stay and install more and more and that the interaction between humans and machines will continue to be conversational, in the form of intelligent assistants that modify our way of working, of creating, of studying, of entertaining, etc., by enhancing and expanding our capabilities and skills, as they can simplify our work by making it more efficient and collaborate in the development of our ideas.

Although, as has happened throughout history in the previous industrial revolutions, certainly some works will disappear, but most will be reformulated and new, so they will also continue to vary the training requirements or skills in different professions and activities, and those who use these tools will have advantages over those who do not.

AI cannot replace us in issues that require empathy, sensitivity, awareness, common sense, active listening, critical thinking, leadership, perception, identification of problems to solve that require knowledge and understanding of a particular cultural or social context, etc.

Just like giving a good prompt ChatGPT is key to generating the outcome we expect, asking permanently good questions remains essential in the face of the new advances and developments that will appear.

These questions should not only focus on what AI can do and how, or whether it will replace us in our work, but also and especially what we want to help or do for us, and what not.

The journey to the future is exciting, dizzy, full of challenges and the AI can be a great copilot in many of the ways to go: let's go!